Introduction
Diabetes mellitus (DM) is a common chronic medical condition affecting approximately 10% of the adult population worldwide [1]. Among those affected, 25%−50% require surgical intervention. This proportion has been steadily increasing as the global burden of DM continues to rise [2,3]. DM is associated with an up to five-fold increase in mortality compared with nondiabetic individuals due to organ damage that worsens with disease progression [4,5]. Diabetic patients have impaired defense mechanisms against surgical stress [6] that further elevates the risk of postoperative complications and ultimately affects recovery and long-term outcomes. Therefore, reducing the incidence of postoperative complications in diabetic patients is a clinical priority in the effort to optimize perioperative management strategies.
Metformin, a first-line treatment for type 2 DM, has been shown to improve glycemic control and has the potential to reduce inflammation and oxidative stress [7]. Given that systemic inflammation contributes significantly to acute organ failure in surgical settings [8], metformin could potentially improve postoperative outcomes via its anti-inflammatory effects. Although this hypothesis has been evaluated in prior studies, the role of metformin in postoperative outcomes remains unclear. The previous investigations conducted in noncardiac surgical settings had several limitations, such as focus on specific types of surgery or short-term follow-up period [9–11]. Moreover, reports on postoperative complications are scarce [12], and the results were heterogeneous throughout the studies [13–15]. Therefore, we aimed to address these gaps by comprehensively examining the associations between preoperative metformin use and postoperative mortality and complications in a large cohort of patients with type 2 DM undergoing noncardiac surgery. We also explored whether different metformin dosages lead to differential outcomes, an issue that has been insufficiently examined in previous research.
Materials and Methods
Study design and population
We conducted a retrospective study using data from our hospital registry that was established to explore risk factors associated with postoperative outcomes after noncardiac surgery. This registry included preoperative, intraoperative, and postoperative data from adult patients (age ≥ 18 years) who underwent noncardiac surgeries from January 2011 to June 2019 at our institution. The Institutional Review Board granted an exemption for this study and waived the need for written informed consent, as all data were collected anonymously.
From this registry, we included patients who were diagnosed with type 2 DM before surgery and were using antidiabetic agents when admitted to hospital. Then, patients were excluded if they met the following criteria: American Society of Anesthesiologists physical status (ASA-PS) classification 5, outpatient surgery, preoperative dialysis, or incomplete medical records. Data were recorded according to Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines [16].
Data collection and potential confounders
All data for this study were obtained by reviewing electronic medical records at our institution using the Clinical Data Warehouse DARWIN-C system that is an electronic data retrieval system designed to enable researchers to extract all types of medical records and laboratory test results in a de-identified form using specific codes or keywords. Mortality data from outside our institution were regularly updated from the National Population Registry of the Korea National Statistical Office using an identification number assigned to each patient.
Two authors independently identified the following potential confounders as covariates: patient characteristics (age, sex, body mass index [BMI], and ASA-PS classification), social and past medical history (alcohol consumption, smoking status, hypertension, chronic kidney disease, liver disease, stroke, peripheral artery disease, coronary artery disease, heart failure, atrial fibrillation, chronic obstructive pulmonary disease, malignancy, and anemia), preoperative medications (calcium channel blockers, angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers, antiplatelet agents, statins, and insulin), and surgery-related factors (emergency, operation duration, intraoperative transfusion, intraoperative inotropic/vasopressor use, type of anesthesia, types of surgery, and risk of surgery). We calculated Charlson Comorbidity Index (CCI) values to evaluate the overall severity of patient comorbidities [17]. Preoperative anemia was defined according to the World Health Organization using the most recent hemoglobin level before surgery [18]. Surgical risks were divided into three groups (low-, intermediate-, and high-risk) according to noncardiac surgery guidelines from the European Society of Cardiology/European Society of Anesthesiology [19].
Exposure and outcome definitions
Preoperative metformin use was defined as use of the drug as an antidiabetic agent for at least 30 days prior to surgery until admission. Pre-admission use of metformin was confirmed by reviewing previous prescriptions and self-medications and verifying these records with pharmacy consultations. Patients taking metformin were divided into low-dose and high-dose groups in accordance with the criteria reported in previous studies [20,21]. If the daily dose was less than 1000 mg, it was defined as a low-dose group, and all higher doses were defined as a high-dose group.
The primary outcome of this study was one-year mortality after surgery. Survival time was calculated from the day of surgery, and each patient was followed until death or September 30, 2024. The secondary outcomes included 30-day mortality, five-year mortality, and the following groups of complications in major organs within 7 days after surgery: cardiac system (myocardial infarction, cardiac arrest), central nervous system (CNS) (stroke, delirium), respiratory system (pneumonia, acute respiratory distress syndrome), or renal system (acute kidney injury). The retrospective diagnosis of interest outcome was identified using relevant International Classification of Diseases 9th/10th Edition codes or clinical definitions presented in Supplementary Table 1.
Statistical analysis
For descriptive statistics, categorical variables were presented as number with percentage and were compared with a χ2 test or Fisher’s exact test. Continuous variables were presented as a median with median (Q1, Q3) and were compared with the Mann–Whitney U test.
To minimize the influence of confounding factors between the two groups, we used stabilized inverse probability of treatment weighting (IPTW). First, we calculated propensity scores using a multivariable logistic regression model that included metformin exposure as the outcome variable and all variables listed in Table 1 as the predictors. Next, the weights were calculated as 1/(probability of exposure to metformin) for patients who had received metformin and as 1/(1 − probability of exposure to metformin) for patients who had not. To avoid the effects of extreme weights, we trimmed all weights less than the first percentile or greater than the 99th percentile by weight truncation [22,23]. The balance between the two groups was evaluated using the absolute standardized differences (ASD) before and after IPTW adjustment, with an ASD less than 0.10 indicative of good balance [24]. The confounding variables that had an insufficient balance (ASD ≥ 0.1) after IPTW were additionally adjusted in the outcome model. To evaluate the dose-related effects of preoperative metformin, we repeatedly generated propensity scores estimating the probability that a patient was exposed to low- or high-dose metformin and calculated weights as previously described.
To compare mortalities between the two groups, Kaplan–Meier survival curves were constructed, and a log-rank test was used to compare the survival distributions of the groups. A Cox proportional hazards regression model was used to calculate the hazard ratio (HR) for mortality, and the proportional hazards assumption was confirmed using a Schoenfeld residual plot by visually assessing whether the residuals were randomly distributed over time. To analyze the effects of metformin on postoperative complications, we used a logistic regression model and reported the results as odds ratio (OR) with 95% CI.
To investigate the heterogeneity of treatment effect of metformin on primary outcome (one-year mortality), we conducted a subgroup analysis using the weighted cohort across the prespecified subgroups: age (< 65 vs. ≥ 65 years), sex (female vs. male), ASA-PS classification (I, II vs. ≥ III), CCI (≤ 2 or > 2), preoperative hypertension (yes or no), risk of surgery (low vs. intermediate to high), and emergent operative status (yes or no). Statistical analysis was performed in R 4.2.0 (http://www.R-project.org), and results were considered statistically significant when the P value was less than 0.05.
Results
Patient characteristics
A total of 22 944 patients were eligible for final analysis after applying inclusion and exclusion criteria from the registry (Fig. 1). Among these patients, 12 536 (54.6%) were exposed to preoperative metformin. The median (Q1, Q3) duration of follow-up after surgery was 8.2 (5.9, 10.9) years. The baseline characteristics of patients according to preoperative metformin exposure are summarized in Table 1. There were some unbalanced baseline characteristics between the two groups, including age, BMI, ASA-PS classification, CCI > 2, anemia, preoperative use of a calcium channel blocker or antiplatelet agent, surgical risk, types of surgery, general anesthesia, intraoperative transfusion, and vasoactive/inotropic use. However, we achieved satisfactory balance (ASD < 0.1) in all variables after IPTW adjustment.
Among the entire 22 944 patients, 5209 (22.7%) received low-dose metformin and 7327 (31.9%) received high-dose metformin. The baseline characteristics of each group compared to the non-metformin group are presented in Supplementary Tables 2 and 3. The ASDs of the covariates in the weighted cohorts were less than 0.1, except for chronic kidney disease (ASD = 0.126), when comparing the low-dose metformin and non-metformin groups.
Primary outcome
The overall mortality within one year after surgery was 5.6% (1275/22 944) in the entire population. The incidence of one-year mortality was 7.4% (766/10 408) in the non-metformin group and 4.1% (509/12 536) in the metformin group (unadjusted HR: 0.54, 95% CI [0.48−0.69], P < 0.001). After weighting, preoperative metformin use was significantly associated with a reduced risk of one-year mortality (HR: 0.76, 95% CI [0.68−0.85], P < 0.001) (Table 2, Fig. 2). In the subgroup analysis, we did not find any significant difference in one-year mortality across the subgroups (Fig. 3). Further, both low- and high-dose metformin groups demonstrated lower hazards of one-year mortality compared with the non-metformin group (HR: 0.70, 95% CI [0.59−0.83], P < 0.001 in the low-dose group and HR: 0.78, 95% CI [0.68−0.90], P < 0.001 in the high-dose group) (Table 3).
Secondary outcome
Before weighting, the 30-day and five-year mortalities after surgery were lower in the metformin group than in the non-metformin group (0.3% vs. 1.0%; P < 0.001 for 30-day mortality and 13.6% vs. 19.7%; P < 0.001 for five-year mortality). The beneficial effects of preoperative metformin use on postoperative mortalities remained significant after IPTW (HR: 0.47, 95% CI [0.31−0.70], P < 0.001 for 30-day mortality and 0.84; 95% CI [0.79−0.90], P < 0.001 for five-year mortality) (Table 2, Fig. 2) and were consistent regardless of the exposure dose (Table 3).
In analyses evaluating the effects of metformin on postoperative organ injury, the metformin group had a lower incidence of all types of complications compared with the non-metformin group before weighting (Table 2). The detailed incidence of each complication is presented in Supplementary Table 4. However, the protective effect of metformin was only valid for the respiratory system and renal system after weighting (OR: 0.76; 95% CI [0.61−0.93], P = 0.01 for respiratory complications and OR: 0.66, 95% CI [0.58−0.74], P < 0.001 for renal complications) (Table 2). In particular, the benefit of metformin on the respiratory system was only evident in the high-dose group (OR: 0.69, 95% CI [0.54−0.89], P = 0.004) (Table 3).
Discussion
We investigated the associations between preoperative use of metformin and postoperative mortality and complications in a large cohort of patients with type 2 DM undergoing noncardiac surgery. Preoperative metformin use was associated with a reduced risk of one-year postoperative mortality regardless of dosage. In addition, metformin was associated with lower odds of postoperative complications in respiratory and renal systems, with the reduction in respiratory complications due primarily to high-dose metformin.
The observed association between metformin use and reduced mortality aligns with previous studies that have shown favorable effects of metformin on survival among diabetic patients undergoing surgery [9,10,15]. However, our study expands on prior research by including a broader spectrum of surgeries and various complications, allowing comprehensive evaluation of the association between preoperative metformin use and postoperative outcomes. Furthermore, by examining dose-response effects, we found that even low-dose metformin was beneficial, though high-dose metformin provided greater protection against postoperative complications.
The protective effects observed in this study may be attributable to metformin’s action mechanisms, including its anti-inflammatory and anti-oxidative properties [25]. These mechanisms may reduce the severity of perioperative inflammatory responses and mitigate the risk of acute organ injury. This is particularly noteworthy for diabetic patients, who are susceptible to postoperative complications that can increase the risk of postoperative morbidity and mortality [26]. Our results suggest that these benefits extend beyond glycemic control, underscoring the potential of metformin to serve as part of an optimal perioperative management strategy aimed at improving postoperative outcomes in diabetic populations undergoing noncardiac surgery.
Nevertheless, metformin showed different effects on postoperative complications, being particularly effective in reducing respiratory and renal complications. This observation suggests that metformin may be more effective in preventing certain types of organ dysfunction, particularly in systems that are highly vulnerable to oxidative stress and inflammation, such as the lungs or kidneys [27,28]. However, we could not find significant reductions in cardiac or CNS complications after weighting. This may be partly explained by differences in pathophysiological mechanisms across organ systems. That is, the lack of an effect on the CNS could reflect the multifactorial causes of complications, and inflammatory modulation alone may not be sufficient for prevention [29,30]. However, these explanations are untested; given the limited and inconsistent reports on postoperative complications in previous studies [13,14], the effects of metformin on specific organ systems require further discussion.
Our study benefits from a large sample size and a broad examination of complications beyond mortality alone. However, it also comes with several limitations. First, as a retrospective analysis, our findings are subject to residual confounding and selection bias, although we used IPTW to mitigate these issues. For example, preoperative hemoglobin A1c levels were measured in only a small subset of patients, preventing us from including this variable as a covariate in our analysis. This limitation may introduce residual confounding, as perioperative glycemic control could influence postoperative outcomes. Second, metformin exposure was determined based on preadmission records that may not reflect compliance rates or actual metformin levels. This reliance on prescription records limits our ability to verify if patients consistently took metformin as prescribed or if there were any dose adjustments prior to surgery. Consequently, variations in adherence or changes in medication regimens might have influenced the outcomes, potentially leading to an underestimate or overestimate of metformin’s true effects on postoperative outcomes. Third, the results may not be generalizable to all patient populations or healthcare settings, as the study was conducted at a single institution. Further prospective studies, ideally randomized controlled trials, are required to provide more robust evidence on the causal relationships between metformin use and postoperative outcomes.
In conclusion, our study suggests that preoperative metformin use in patients with type 2 DM undergoing noncardiac surgery is associated with reduced one-year mortality and a lower incidence of respiratory and renal complications. These findings indicate the potential of metformin as an adjunctive strategy in the perioperative care of diabetic patients. Further research is needed to explore the mechanisms underlying these benefits and to determine the optimal dosing strategies to maximize metformin’s protective effects in surgical settings.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement1
Print
